You can use codes we have adapted from the text Numerical Recipes in C++, see chapter 10.7. Here we present a program, which you also can find at the webpage of the course we use the functions dfpmin and lnsrch. This is a variant of the Broyden et al algorithm discussed in the previous slide.

• The program uses the harmonic oscillator in one dimensions as example.
• The program does not use armadillo to handle vectors and matrices, but employs rather my own vector-matrix class. These auxiliary functions, and the main program model.cpp can all be found under the program link here.

Below we show only excerpts from the main program. For the full program, see the above link.